Diffusion tensor imaging and diffusion kurtosis imaging of the pancreas - feasibility, robustness and protocol comparison in a healthy population

• Published in: Abdominal radiology (New York)
• Main Authors: Bilreiro, Carlos, Andrade, Luisa, Henriques, Rafael, Loução, Nuno, Matos, Celso, Shemesh, Noam
• Format: Journal Article
• Language: English
• Published: United States 26.03.2025
• Subjects: Healthy volunteers; Pancreas; Diffusion kurtosis imaging; Feasibility studies; Abdomen; Diffusion tensor imaging
• ISSN: 2366-0058; 2366-0058
• DOI: 10.1007/s00261-025-04889-w

This study aims to determine the feasibility, image quality, intra-subject repeatability and inter-reader variability of Diffusion tensor (DTI) and Diffusion kurtosis imaging (DKI) for pancreatic imaging using different protocols and report normative values in healthy individuals. Single-institution prospective study performed on healthy volunteers in a clinical 3T scanner, using two different protocols (6/16 diffusion directions). Acquisitions were repeated twice to assess intra-subject repeatability. To assess inter-reader variability, Mean diffusivity (MD), Axial diffusivity (AD), Radial diffusivity (RD), Apparent diffusion coefficient (ADC) and Mean kurtosis (MK) values were extracted from segmented pancreas by two radiologists. A Likert scale was used by both readers to assess subjective image quality. Twelve healthy volunteers were recruited for each MRI protocol. The 6 diffusion directions protocol was shorter: 7 min vs. 14 min (corresponding to 4 min vs. 7.5 min for a DTI only reconstruction). No differences in image quality were found between protocols. Only MK maps showed implausible estimates, leading to the exclusion of median 16% and 17.7% pixels for the 6- and 16-direction protocols, respectively. Intra-subject repeatability was determined with negligible coefficients of repeatability for DTI; however, MK presented slightly higher values. Inter-reader agreement was excellent for all maps (ICC > 0.9). DTI and DKI of the pancreas are feasible in clinical settings, with excellent inter-observer agreement and good image quality. Intra-subject repeatability is excellent for DTI, but some variability was observed with DKI. A 6-directions protocol may be preferred due to faster acquisition without quantitatively compromising estimates. MK inaccuracies prompt further research for improving artifact correction.